Desertification detection model in Naiman Banner based on the albedo-modified soil adjusted vegetation index feature space using the Landsat8 OLI images

Abstract

The current desertification feature space models are almost linear and ignore the complicated and nonlinear relations that exist among variables when monitoring desertification. Herein, point-to-point and point-to-line models have been proposed by completely considering the nonlinear relations between the Albedo-Modified Soil Adjusted Vegetation Index (MSAVI) and the effects of soil background. Further, the applicability of these models for monitoring different levels of desertification information was compared and analyzed. The point-to-line feature space model exhibited a larger inversion accuracy (93.8%) for Naiman Banner with respect to albedo-MSAVI when compared with that exhibited by the point-to-point model (88.9%). In addition, the monitoring accuracy is observed to slightly differ for different levels of desertification, and slight and mild desertification exhibit the best inversion accuracy in case of both point-to-point and point-to-line models. Furthermore, the point-to-line model exhibits better applicability in case of intensive (92.7%) and severe (93.3%) desertification when compared with those exhibited by the point-to-point model (87.5% and 88.9%, respectively). The results obtained in this study can provide improved data and decision support for preventing and managing land degradation.